{"id":"cefbb958-0193-4ae5-a459-b7c551fb805e","arxiv_id":"1908.03951","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Velocity structure functions in simulated molecular clouds show turbulence at early times and gravitational collapse later, both of which can reproduce Larson's size-velocity relation.","lead":"Using supercomputer simulations of galaxy gas, this study tracks how velocity structure functions, a statistical measure of gas speed differences, change as three molecular clouds evolve. It shows that the classic Larson relation between cloud size and gas speed can arise from two different physical causes at different times, turbulence early on and gravitational collapse later.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The late-time VSF flattening/inversion central to the collapse diagnostic is only measured at λJ=4Δx; the sole higher-resolution test ends before the contraction phase and shows up to 100% changes in ζ(3), so the main conclusion is not numerically converged.","rationale":"I read the paper as a careful, transparent numerical study whose central claim is that the evolution of density-weighted VSF slopes—especially the late-time flattening/inversion of ζ(p)—can distinguish gravitational contraction from turbulence, while the global σ–R relation cannot. The analysis is well-documented, the parameter study (1D/3D, threshold, weighting, refinement) is valuable, and the public data and scripts are a real strength. My main concern is the same one the reader identified: numerical convergence of the very exponents that carry the conclusion. The paper's own refinement comparison is limited to M3 and to t ≤ 1.2 Myr, and it shows large differences (40% in ζ(2), 100% in ζ(3)) at λJ=32Δx. The authors interpret these as a shock-resolved-at-high-resolution effect; even if that interpretation is correct, it demonstrates that higher-order ζ values are resolution-sensitive in this regime, leaving the late-time contraction signal without a convergence check. The λJ=8 run agrees with λJ=4 but is not an adequate convergence test because it still misses 13% of kinetic energy and was not compared to λJ=32 beyond 1.2 Myr. I also note that the Appendix A weighting function's 8 pc break is not sensitivity-tested, which adds an avoidable systematic uncertainty to all quoted exponents. None of this proves the headline claim wrong—indeed the weighted/unweighted contrast and the extended self-similarity analysis support a physical interpretation—but it means the central diagnostic is conditional on resolution. Hence I do not change the reader's CONDITIONAL verdict.","tokens_in":25990,"tokens_out":6812,"duration_ms":72641,"concrete_test":"Extend the M3 λJ=32Δx simulation through the contraction phase (to t ≥ 3 Myr after self-gravity onset, the interval where fiducial ζ(3) crosses zero), recomputing density-weighted ζ(1), ζ(2), ζ(3) with the same 0.8–8 pc fitting range and Appendix A weighting. If ζ(p) still drops below ~0.5 or inverts at late times, the diagnostic is robust; if the drop/inversion weakens or disappears, the central claim is an artifact of the 4Δx Jeans criterion. As a cheaper secondary check, re-fit the existing public λJ=4Δx VSF data with several alternative weightings (uniform to 8 pc, no weighting, different break radii) to bound the systematic error in ζ from the ad hoc 8 pc break.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central diagnostic—density-weighted ζ(p) flattening or inverting as gravitational contraction proceeds—is measured exclusively in the fiducial λJ=4Δx runs (Fig. 2a, Sect. 3.2). The only higher-resolution check (Sect. 3.6, Fig. 4) covers M3 for just the first 1.2 Myr: at t=0.8 Myr, ζ(2) and ζ(3) are 40% and 100% higher at λJ=32Δx than at λJ=4Δx. The authors attribute this to an SN shock only the high-refinement run captures, but the same plot shows that ζ(p) is resolution-sensitive exactly in the phase used to establish the 'uniform turbulence reproduces Larson' part of the claim. More importantly, the late-time contraction phase (t ≳ 2–3 Myr) where ζ(3) approaches zero or becomes negative has no λJ=32 counterpart; the λJ=8 run agrees with λJ=4 but misses 13% of the kinetic energy and is itself not a convergence proof. The fitted exponents also depend on the Appendix A weighting that downweights lags above 8 pc (w = 1 pc/l), with no sensitivity test. Until the late-time inversion is shown at higher refinement, the paper's headline claim—that VSF exponent drop/inversion is a diagnostic of gravitational contraction—is not numerically established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper analyzes velocity structure functions (VSFs) of three molecular clouds formed in the magnetohydrodynamic simulations of Paper I/II, using density-weighted and unweighted VSFs, with and without density thresholds, in 3D and projected 1D, and at different Jeans refinement levels. The authors measure the scaling exponents ζ(p) for p = 1–3 from power-law fits over 0.8–8 pc (with an additional weighting function beyond 8 pc), as well as the extended self-similarity ratio Z(p) = ζ(p)/ζ(3). They report that at early times the density-weighted VSFs are consistent with supersonic turbulence and Larson's size–velocity relation; SN blast waves produce transient increases in ζ; and during gravitational contraction the density-weighted ζ(p) drops and can become negative as small-scale motions dominate, while the unweighted VSF and the global velocity–radius relation remain positive and Larson-like. They conclude that the turbulent-cascade and gravitational-collapse explanations of Larson's relation coincide by accident at different evolutionary stages, and that VSF flattening/inversion is a potential diagnostic of collapse.","tokens_in":26254,"tokens_out":7999,"duration_ms":80460,"significance":"If the central claim is correct, it offers a practical discriminator between turbulence-dominated and collapse-dominated molecular cloud kinematics, addressing a long-standing ambiguity in the interpretation of Larson's relation. The paper also provides useful methodological comparisons of 1D versus 3D VSFs, density thresholds, and density weighting, and it ships the simulation data and scripts in a public repository. The analysis is clearly described, the statistical fits are characterized with error estimates, and the observational comparison is reasonable. However, the headline conclusion is conditional on numerical convergence of the density-weighted ζ(p) during the contraction phase, which the manuscript does not currently establish.","major_comments":[{"comment":"The late-time flattening/inversion of the density-weighted ζ(p), which is the paper's proposed collapse diagnostic (Sect. 3.2, Fig. 2a), is measured only in the fiducial λJ = 4Δx runs. The only higher-refinement runs for M3 extend only to t = 1.2 Myr (λJ = 32Δx) and t ≈ 3 Myr (λJ = 8Δx). At t = 0.8 Myr, ζ(2) and ζ(3) are 40% and 100% higher at λJ = 32Δx than at λJ = 4Δx (Fig. 4, bottom). Even if this excursion is attributed to an SN blast wave that only the high-refinement run captures, the comparison demonstrates that ζ(p) is strongly resolution-dependent in the early phase used to establish the 'uniform turbulence' part of the argument, and there is no higher-refinement coverage of the t ≳ 2–3 Myr contraction phase in which ζ(3) approaches zero or becomes negative. Since Paper II reports that λJ = 4Δx misses 13–33% of the kinetic energy, the headline claim that VSF drop/inversion is a diagnostic of gravitational contraction is not numerically converged. The authors should either extend a high-refinement run through the contraction phase or explicitly present the collapse diagnostic as tentative and resolution-dependent.","section":"Sect. 3.6, Fig. 4"},{"comment":"The fitted exponents ζ(p) depend on the ad hoc weighting function w(ℓ) = 1 for 0.8 ≤ ℓ ≤ 8 pc and w(ℓ) = 1 pc/ℓ beyond 8 pc. No sensitivity test is reported, although Sect. B.1 emphasizes that the VSFs are often not single power laws but superpositions of several driving processes. The quantitative values, including the sign crossovers that drive the collapse interpretation, may therefore depend on the chosen break scale and on the 8–30 pc weighting. Please provide fits with alternative weighting schemes (e.g., uniform weighting over the full fitting range, or different break scales) or demonstrate directly from the unweighted S_p(ℓ) curves that the flattening/inversion conclusion is robust.","section":"Appendix A, Eq. (A.1)"}],"minor_comments":[{"comment":"The caption lists the three panels as '(left to right) λ = 4 Δx, λ = 8 Δx, and λ = 8 Δx'; the third should read λ = 32 Δx.","section":"Appendix B.3, Fig. B.4 caption"},{"comment":"The sentence 'our findings are generally consistent with with observations' contains a duplicated 'with'.","section":"Sect. 5, item 8"},{"comment":"The word 'proibitively' should be 'prohibitively'.","section":"Sect. 4.3"},{"comment":"The phrase 'Paper II discuss' should be 'Paper II discusses'.","section":"Sect. 3.2"},{"comment":"The sentence 'the values of ζ are about four times steeper than the values that are predicted by (Boldyrev 2002) for incompressible flows' should presumably say 'for compressible/supersonic flows', since Boldyrev (2002) treats supersonic turbulence.","section":"Sect. 3.4"},{"comment":"The statement that extended self-similarity 'removes the effects of hierarchical gravitational collapse' should be restricted to epochs where ζ(3) is not close to zero, as the authors themselves note that the procedure fails at zero crossings; this caveat should appear in the summary as well.","section":"Sect. 4.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the scope of the journal and the methodological comparisons are valuable, especially the public release of data and scripts. The main obstacle is that the central collapse diagnostic is not numerically converged at the fiducial resolution; the authors either need to supply the decisive late-time high-resolution run or reframe the headline claim as a resolution-dependent tendency. I would not reject, because the raw qualitative behavior is suggestive and the paper is otherwise carefully presented."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a transparent and mostly careful VSF study, and the central idea is worth taking seriously, but the collapse diagnostic that drives the conclusions is not numerically converged in the regime where it matters.\n\nThe genuinely new thing here is the time-resolved comparison of density-weighted versus unweighted VSFs in self-consistently formed cloud simulations. Showing that density weighting selects for collapse signatures while the global velocity-size relation still follows Larson's relation is a real step beyond Paper I and earlier work by Kritsuk et al. The paper also does several things well: the method section is clear, the parameter study (1D vs 3D, density threshold, weighting, Jeans refinement) is thorough, the fit errors are reported, and the data and scripts are public. The authors are also upfront about the kinetic energy missing in the fiducial runs.\n\nThe soft spot is exactly where the stress-test note lands. The late-time flattening and inversion of the density-weighted ζ(p) — the paper's headline diagnostic — is only measured at λJ = 4Δx. The single higher-resolution check (λJ = 32Δx) covers only the first 1.2 Myr, and there ζ(2) and ζ(3) are 40% and 100% higher than in the fiducial run. The authors attribute that to a SN shock that only the high-resolution run captures. That may be right, but it means the resolution sensitivity is demonstrated precisely in the phase used to establish the 'uniform turbulence reproduces Larson' half of the claim, and the contraction phase itself has no high-resolution counterpart. The λJ = 8 run agrees with λJ = 4 but misses 13% of the kinetic energy and is not a convergence proof. So the paper's own statement in Sect. 4.5 that 'the overall behaviour is already determined by our moderate resolution simulations' is stronger than the evidence supports. I'd call this a major comment, not a fatal flaw: the qualitative picture is plausible and consistent with physical expectations, but the exponent values should be treated as provisional.\n\nA minor issue: the power-law fits use an ad hoc weighting break at 8 pc with no sensitivity test (they note this in Appendix A). And the observational comparison is honest but limited by the lack of radiative transfer, which they acknowledge.\n\nWho this is for: observers working on linewidth-size relations and anyone testing diagnostics of gravitational collapse versus turbulence in simulations. It deserves a serious referee: the analysis is transparent, the data are public, and the conclusion, if it survives the convergence test, would be useful. I'd send it to review with a request for a higher-resolution late-time run or a much stronger caveat.","headline":"A transparent and useful VSF study of simulated clouds whose central collapse diagnostic needs one more convergence test before the exponent values can be trusted.","tokens_in":26849,"tokens_out":2988,"would_cite":true,"duration_ms":30157,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes that the density-weighted velocity structure function can distinguish whether a molecular cloud's observed size–velocity relation is produced by a turbulent energy cascade or by gravitational contraction.","keywords":["velocity structure functions","molecular clouds","Larson size-velocity relation","supersonic turbulence","gravitational collapse","supernova blast waves","density weighting","adaptive mesh refinement simulations"],"falsifier":"Re-run the same three clouds with the Jeans length resolved by 32 grid cells for the full evolution, and watch the density-weighted exponents through the contraction phase; if $\\zeta(3)$ no longer drops toward zero or negative, the collapse signature in the fiducial runs is an artifact of under-resolved turbulence. Observationally, a high-resolution study of an actively star-forming cloud with an optically thin dense-gas tracer should find density-weighted $\\zeta(1)$ approaching zero or going negative if the inversion is a real property of collapsing gas.","tokens_in":25712,"feed_emoji":"🌌","tokens_out":8491,"duration_ms":83413,"temperature":0.7,"pith_summary":"This paper argues that the familiar Larson size–velocity relation of molecular clouds is not one effect but two. In three clouds that form self-consistently in kiloparsec-scale simulations of the interstellar medium, the density-weighted velocity structure functions initially show the power-law scaling expected from a turbulent energy cascade, with first-order exponent $\\zeta(1)\\simeq 1/2$. Once self-gravity begins to dominate, the same global relation $\\sigma\\propto R^{1/2}$ persists—now reflecting virial equilibrium or free-fall collapse—while the density-weighted exponents $\\zeta(p)$ become shallow or even negative as kinetic energy concentrates on small scales. The paper concludes that the velocity structure function, not the global size–velocity relation, is the diagnostic that tells these two regimes apart. It also finds that supernova blast waves temporarily steepen the exponents for about a crossing time, so shock-impacted clouds can briefly masquerade as more turbulent than they are.","feed_headline":"Velocity structure functions separate turbulence from collapse","feed_subtitle":"Clouds reach the same size–velocity line for two different reasons; density-weighted scaling shows which.","key_machinery":"The load-bearing object is the density-weighted velocity structure function, a two-point correlation of velocity differences $\\Delta v$ between gas elements separated by lag $\\ell$, with pairs weighted by the density product $\\rho(\\mathbf{x})\\rho(\\mathbf{x}+\\boldsymbol{\\ell})$. Fitting $S_p(\\ell)\\propto \\ell^{\\zeta(p)}$ over lags of 0.8–8 pc yields the scaling exponents $\\zeta(1),\\zeta(2),\\zeta(3)$; the self-similarity ratio $Z(p)=\\zeta(p)/\\zeta(3)$ measures the shape independent of amplitude. The mechanism is that density weighting emphasizes the dense gas where collapse acts, so the decline and inversion of $\\zeta(p)$ signal gravitational contraction, while $Z(p)$ continues to match compressible-turbulence predictions unless a shock or a zero-crossing of $\\zeta(3)$ intervenes.","core_discovery":"The central claim is that a single power-law fit to the density-weighted velocity structure function $S_p(\\ell)=\\langle\\rho(\\mathbf{x})\\rho(\\mathbf{x}+\\boldsymbol{\\ell})|\\Delta v|^p\\rangle/\\langle\\rho(\\mathbf{x})\\rho(\\mathbf{x}+\\boldsymbol{\\ell})\\rangle$ tracks which physical process dominates a molecular cloud's internal motions. Early in the simulated evolution, uniform turbulence dominates and the fitted exponents sit near the compressible-turbulence predictions, with $\\zeta(1)\\simeq 0.5$; later, gravitational contraction transfers velocity power to small scales and the density-weighted $\\zeta(p)$ drop, in some cases below zero. Remarkably, the cloud's global linewidth-radius relation remains $\\sigma\\propto R^{1/2}$ throughout, because free-fall and virial equilibrium produce the same functional form. The authors therefore claim that two different mechanisms coincidentally generate Larson's relation, and that only the two-point statistics of the velocity field reveal which mechanism is at work.","pith_inferences":["If the collapse signature survives at higher resolution, the density-weighted VSF exponent could serve as an observational 'collapse meter' that ranks clouds by evolutionary stage in large surveys without full radiative-transfer modeling.","The paper's logic implies that scatter around Larson's relation is not pure noise: clouds in the turbulence-dominated stage and clouds in the collapse stage both sit on the relation, so residuals may encode which regime dominates.","Because CO becomes optically thick in the densest gas, existing observed VSF exponents may be biased toward the turbulent-envelope values; optically thin high-density tracers should reveal flatter exponents in actively star-forming regions.","The persistence of extended self-similarity during collapse suggests that $Z(p)$ decouples from gravity and measures the turbulent cascade alone, a separation that future analyses could exploit to isolate turbulence from collapse in observations."],"forward_implications":["A measured VSF slope can be read as an evolutionary stage: $\\zeta(1)\\simeq 1/2$ points to turbulence-dominated gas, while shallow or negative density-weighted slopes signal ongoing gravitational contraction.","The global $\\sigma\\propto R^{1/2}$ relation alone is degenerate, so observers who rely on it cannot determine whether turbulent support or collapse dominates a cloud.","Supernova shock impacts imitate extra turbulence for roughly a crossing time, so surveys that catch post-impact clouds may overestimate turbulent driving.","One-dimensional line-of-sight VSFs reproduce the three-dimensional behaviour in most cases, supporting the use of CO-based VSFs as dynamical diagnostics in bound, contracting clouds.","Analysis choices such as density thresholds and density weighting systematically change the fitted exponents, so observational comparisons must match those choices to the tracer's density coverage."],"supporting_citations":[{"why":"Supplies the kpc-scale ISM simulations in which the clouds form and the earlier finding that uniform turbulence alone cannot reproduce the observed relation.","marker":"Paper I"},{"why":"Defines the three high-resolution clouds and reports the Jeans-refinement tests showing kinetic-energy shortfalls at $\\lambda_J=4\\Delta x$.","marker":"Paper II"},{"why":"Presents the size-velocity relation that the paper sets out to explain.","marker":"Larson (1981)"},{"why":"Gives the incompressible-turbulence VSF scaling used as one baseline.","marker":"She & Lévêque (1994)"},{"why":"Gives the supersonic-compressible VSF scaling used as the second baseline.","marker":"Boldyrev (2002)"},{"why":"Introduces the density-weighted VSF definition adopted throughout the analysis.","marker":"Padoan et al. (2016b)"},{"why":"Shows that free-fall and virial equilibrium both yield a square-root size-velocity relation, grounding the collapse-stage interpretation.","marker":"Ballesteros-Paredes et al. (2011c)"}],"fun_headline_variants":["Velocity structure functions expose turbulence vs collapse","Same Larson line, different cause: structure functions tell","Structure functions unmix turbulence and gravitational collapse","Density-weighted scaling separates collapse from turbulence","How clouds hide collapse: structure functions reveal it"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results rest on the assumption that resolving each Jeans length with only four grid cells still captures the turbulent cascade in dense gas well enough that the reported exponents, especially $\\zeta(2)$ and $\\zeta(3)$, reflect real physics; the paper's own convergence test shows $\\zeta(3)$ running 40–100 percent higher when the Jeans length is resolved with 32 cells, so the higher-order exponents are not numerically settled.","fun_headline_variants_meta":{"raw":{"variants":["Velocity structure functions expose turbulence vs collapse","Same Larson line, different cause: structure functions tell","Structure functions unmix turbulence and gravitational collapse","Density-weighted scaling separates collapse from turbulence","How clouds hide collapse: structure functions reveal it"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000232,"raw_usage":{"total_tokens":1543,"prompt_tokens":1051,"completion_tokens":492,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":667,"completion_tokens_details":{"reasoning_tokens":424}},"tokens_in":667,"tokens_out":492,"duration_ms":5695,"temperature":1.0,"reasoning_tokens":424,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:56:31.432156+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the same three clouds with the Jeans length resolved by 32 grid cells for the full evolution, and watch the density-weighted exponents through the contraction phase; if $\\zeta(3)$ no longer drops toward zero or negative, the collapse signature in the fiducial runs is an artifact of under-resolved turbulence. Observationally, a high-resolution study of an actively star-forming cloud with an optically thin dense-gas tracer should find density-weighted $\\zeta(1)$ approaching zero or going negative if the inversion is a real property of collapsing gas.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Presents the size-velocity relation that the paper sets out to explain."},{"cited_title":"& L´evˆeque, E","cited_arxiv_id":null,"evidence_quote":"Gives the incompressible-turbulence VSF scaling used as one baseline."},{"cited_title":"2002, Astrophys","cited_arxiv_id":null,"evidence_quote":"Gives the supersonic-compressible VSF scaling used as the second baseline."}],"review_version":1}